[GRASSLIST:3451] Re: seeking advice on improving results with s.surf.rst

[Steve Trick]

On 25 Mar 2002 at 12:02, Neil Ripley wrote:

> The problem I have is that the segment edges are very apparent in the
> results I am getting from s.surf.rst. I am not getting a smoothly
> continuous surface as I expect I should. Instead I have clear and
> distinct breaks that coincide with the segmentation boundaries.

Hi Neil,

I had similar problems getting good interpolation results using 
contour lines data.

Better results were achieved using the following method (based 
on advice from Helena Mitasova).

Create an interpolated surface from your contour data, using 
s.surf.rst, with the default parameters.  Then run r.random on 
this surface to create some additional 'spot heights'.  Then add 
those random points to your original sites file that you used to 
generate the first surface.  Then run s.surf.rst again.  Since 
s.surf.rst is more used to using distributed point data, you 
should get better results.

Even though this method worked for me, I decided to try some 
alternative routes.  I managed to get some very good results 
using the Surfer program, which I was fortunate to have access 
to.  I exported my GRASS sites file to an ascii file that Surfer 
could read.  Surfer can generate interpolated sufaces using a 
wide range of methods.  I found the 'natural neighbour' method 
worked best with contour data.  You can then import the Surfer 
raster back into GRASS using r.in.ascii, using the -s switch.

Hope this helps

Steve
-----------------------------------
Steven Trick BA MSc
School of History and Archaeology
Cardiff University
PO Box 909
Cardiff
Wales
CF10 3XU
email: TrickSG1 at cf.ac.uk
mob: 07799897107
http://www.cardiff.ac.uk/hisar/people/st/

------ Petition for a Software Patent Free Europe:
http://petition.eurolinux.org/index_html?LANG=en